Neither you nor I have consented to the genes we inherited. Our general makeup was dictated to us. I had no say whatever in my allergies, my short stature, my baldness, my tone deafness, nor in my general optimistic bent, my blue eyes, and my affinity for chocolate. Neither did my parents have much say in my genetic makeup, beyond their choice to birth me. My genes were picked at random, and so I accept all these traits as a given. As has every generation before me. We might clench our fists and rail against the gods for our genetic lot — as I am sure I would do if I had a serious genetic disability — but we can not be mad at our ancestors. Everyone has inherited at least one fatal genetic disease, a disease we all share — called Old Age Death.

We can accept the whims of non-consensual random picks of traits in our mortal body as the best we can do for ourselves, or we can bring evidence and rational scientific method and try to do better by selecting genes and traits. However, long before genetic programming becomes common in humans it will be tested and perfected in animals. The economics of animal breeding are huge and probably greater than the market for engineering in humans in the beginning, so until crispr and other techniques are used in animal breeding on a routine basis to work out the kinks, they are not going to be widespread in humans. This means that we can accumulate a lot of evidence of what happens when we start engineering organisms. And we already have a lot of evidence of what happens in some animals like mice.

One of the principles of animal breeding is that traits are not items that we can simply check off from a shopping list and download into the organism. Traits don’t work that way. Yes, it is possible to insert a list of genes, but that does not mean you’ll get a string of traits. Most traits are not generated by a single gene, but rather by a complex of genes. More importantly, even single gene traits are modified, offset, altered, or displaced by the action of other genes. A trait emerges from an ecosystem of interacting genes, all influencing each other. This complexity is a challenge for science.

Let’s say I make up a list of all the genes — or traits — I would like to have in my children. Super smart, able to run a marathon, no fear of heights, risk taking, extrovert, compassionate, photographic memory, empathetic, tall, thin, sonorous voice, easy going, no allergies, comfortable with math, good listener. Who would not want all these in their children?

The challenge is the human proteins needed to make each of these to happen can conflict with each other. The proteins — created by the genes — needed for risk taking may be the ones that dampen good listening. There are genetic trade offs, in that you cannot optimize all traits. In other words there are genetic costs for each trait. To raise IQ might cost lowering something else, such as empathy. It’s not that there is a zero-sum quantity being conserved, it’s that genes cannot do all that is possible. They are hugely constrained by each other. It is like trying to design a machine: it cannot optimize all properties; it cannot be fastest, lightest, strongest, and cheapest at the same time. Everything is a trade off. Performance, reliability, speed, cost — all are trade off between them. Genes and traits also operate under the same regime of trade offs. There are no free lunches.

However there are some genetic diseases that many people fear in their children that are so horrible that they may be willing to pay any genetic cost to have them eliminated as a possibility. These are likely to be the first kinds of genetic engineering we do in our children.

Scientifically we are nowhere near being able to insert or delete genes at will into the next generation of humans. That is still a far dream. The current method closest to that goal is embryo selection. Parents use IVF to make many embryos, then genetically sequence all of them, and then based on the results and their preferences for certain traits with genetic ties, select those embryos with the highest or lowest scores on those traits. There are several companies already offering this service to couples, and already babies born via this process.

As you might imagine, this is a very fuzzy process. First, we don’t have very much scientific evidence for genetic links to most traits, so the list of things we can choose is small. Secondly, all the embryos are siblings, and so the trait variance between them is not that huge, maybe only a few percent difference. And lastly, we don’t really have a clue about the trade-offs in the presence of desirable traits and absences of undesirable traits.

In addition to our list of desirable genetic traits, we’d also like to have a list of their corresponding genetic costs. And in parallel, for every undesirable trait or gene we want to get rid of, we should know what its corresponding genetic costs are for removing it. When we are aware of genetic side effects, by products, unintended consequences, and natural trade offs of traits, this knowledge makes choosing traits much more difficult. In fact some parents will decide to rely on random chance (sex!) rather than make those choices and feel that weight of responsibility. Others will wait until we have a lot more certainty for what the genetic costs are for popular desirable traits.

We can speed this knowledge up by increasing the extent of our animal models, and we can also use the new techniques of computer simulations of genetics and AI models to further increase our understanding of gene expression. But the real proof — and the only proof acceptable for many parents — will be what happens when selected embryos grow up to adulthood, and that is a very slow process that can not realistically be sped up.

Replacing our reliance on random choice with informed choice for next generations is inevitable, and a net good. But genetic choice will be a slow moving frontier due to the difficult science, the inherent slow process of human growing-up, and the reluctance of parents to feel responsible for genetic consequences. All these will be headwinds, in addition to the expected societal resistance to anything that smells like eugenics, or is brand new.

Slow is good for genetic choice. It gives us time to evaluate our experiments, to deepen our understanding, to develop better tools, and to educate ourselves in the nature our responsibilities to future generations.